Synthetic Spotting
A discrete user judgement that a given piece of content was produced by an AI system, inferred from stylistic regularities, characteristic phrasing, or content-level cues. The referent is the act of attribution, not the correctness of that attribution.
Operational Definition
EVENT. One occurrence is recorded each time a user explicitly attributes a presented artefact (text, image, or audio) to AI generation. The observable marker is a verbal or logged attribution judgement ('this is AI-generated'). When ground truth of the artefact is known, each judgement is scored hit / miss / false alarm / correct rejection.
Measurement Schema
Proposed measurement protocol (not yet empirically validated): Signal-detection task: participants classify a balanced stimulus set of AI-generated and human-authored artefacts. Sensitivity could be reported as d-prime and response bias as criterion c, computed from hit and false-alarm rates; per-stimulus confidence captured for an ROC curve. Accuracy contrasted against a 50% chance baseline with a binomial test.